Improving Model-Based Gas Turbine Fault Diagnosis Using Multi-Operating Point Method

Abstract

A comprehensive gas turbine fault diagnosis system has been designed using a full nonlinear simulator developed in Turbotec company for the V94.2 industrial gas turbine manufactured by Siemens AG. The methods used for detection and isolation of faulty components are gas path analysis (GPA) and extended Kalman filter (EKF). In this paper, the main health parameter degradations namely efficiency and flow capacity of the compressor and turbine sections are estimated and the responsible physical faults such as fouling and erosion are found. Two approaches are tested: The single-operating point and the multi-operating point. Simulation results show good estimations for diagnosis of most of the important degradations in the compressor and turbine sections for the single-point and improved estimations for the multi-point approach.